“If we continue to develop our technology without wisdom or prudence, our servant may prove to be our executioner.”
– Omar N. Bradley
Three Types of Slaves
Slaves in Ancient Greece were pretty close to our contemporary idea of a slave. Wikipedia says, “The condition of slaves varied very much according to their status; the mine slaves of Laureion and the brothel prostitutes lived a particularly brutal existence, while public slaves, craftsmen, tradesmen and bankers enjoyed relative independence.” Slave bankers, what a concept! Often, the work that slaves produced was entirely for the benefit of their owners. This meant that slaves had no positive incentive to work hard, and motivation was often applied in the form of punishments rather than rewards.
Janissaries in the Ottoman Empire were different. Scouts from the Empire would visit villages and pick out the fittest teenage boys, who would be taken away from their families and trained to be ruthless and skilled soldiers, like the Unsullied in Game of Thrones. Surprisingly, they led lives of luxury outside of wars and could own property, had high social status, and in many respects had better lives than ordinary citizens. It says here, “the classification of a slave in medieval Islamic society is rather misleading, if perceived through the lens of our modern sensibility. In that regard, in the Ottoman scheme of things, a slave or a Kul often enjoyed more social benefits and even better opportunities than an ordinary subject (as opposed to oppression), at least till the 17th century. Essentially, they displayed their ‘slave’ status with pride.” Thus, there was significant incentive to be good at their job, and punishment was rarely used to motivate them.
Japanese prisoners of war in World War 2 were slave laborers. For example, while building the Death Railway, “Between 180,000 and 250,000 Southeast Asian civilian labourers and about 61,000 Allied prisoners of war were subjected to forced labour during its construction. About 90,000 civilian labourers and more than 12,000 Allied prisoners died.” This railway transported war materiel for the Empire of Japan. For the prisoners, working harder harmed them more, because it increased the fighting strength of their enslavers. Unsurprisingly, the incentives for prisoners to work were punishment-based, and the low survival rates reflected this.
* * *
In our complacent developed world, the Ancient Greek form of slavery is rare. The closest analog may be low-skilled retail workers. Flipping burgers and taking orders more quickly makes more money for McDonald’s, but won’t have an effect on the employee’s wages. That’s why we see adversarial labor relations, and explains the restrictive rules and restrictions that managers place on the employees.
White-collar workers, especially senior ones, are more like Janissaries. The company reaps the majority of the benefit of their work. Unlike fast-food workers, though, these managerial and professional classes are compensated well, and can capture some of their output for themselves. Building up personal networks is a completely legal way to do so. Grabbing perquisites is in a grey area, while fraud and embezzlement are illegal options available to some. That’s why these workers are motivated mainly by propaganda, sloganeering, social pressure, and status.
At first glance, it looks like there is no modern parallel to Japanese prisoners of war – no one is literally enslaved, so no one would work to harm themselves. But what about those hiding in plain sight: the techno-optimists?
Techno-optimists believe that technology is a force for good in the world because it increases economic prosperity. Technology is the primary determinant of Total Factor Productivity, and productivity determines prosperity in the long run.
The classic objection to seeing technology as an unalloyed good is that technology is a tool, and more powerful tools intensify outcomes. Nuclear fission can cause great destruction in the form of nuclear bombs, but can solve the problem of climate change if used to power nuclear reactors. Although nuclear energy might bring benefits, these benefits aren’t worth the chance of catastrophe. In expected value terms, the technology is negative. Many technologies are double-edged swords, and humans are too irresponsible to be trusted to use them wisely. Thus, it’s better to leave these technologies alone.
I share the techno-optimists’ dismissal of this objection. Humanity is able to handle these dangers. For example, most terrorist attacks use low-tech weapons, even though the technology for more lethal weapons is public information. Scott Adams, a cartoonist, says that “whenever humanity can see a slow-moving disaster coming, we find a way to avoid it,” and I agree with him.
Instead, I see a danger that’s invisible to the irrepressible techno-optimists. It’s the danger that people will be making the world worse when they think they are making it better.
Stubborn Attachments, a book by economist Tyler Cowen, argues that we should maximize economic growth. Economic growth is the most important determinant of human welfare, and it compounds exponentially. Other important things don’t compound exponentially, so we should rank economic growth above other worthy causes such as inequality. This principle is, “We should make political choices so as to maximize the rate of sustainable economic growth.”
Technology is pretty much the only factor which compounds exponentially, and so it follows that increasing the rate of technological advancement should be the highest priority of a society.
The economic growth principle has a single exception: “We should push for sustainable economic growth, but not at the expense of inviolable human rights.” This is a jarring contrast to the rest of the book, which lauds the many benefits of economic growth. In fact, it is the only exception to his rule of prioritizing economic growth: “If the time horizon is sufficiently long, the only non-growth-related values that will bind practical decisions are the absolute side constraints, or the inviolable human rights.”
What are these rights? Tyler gives as an example, “we shouldn’t murder a baby to increase national income by $5 billion.” The right not to be murdered is sacrosanct, even though the world would be more prosperous, because it’s an inviolable human right.
The book doesn’t define the contours of these inviolable human rights, so I’ll suggest a more controversial example.
Imagine an alternate reality earth which is the same as ours, but with one difference: France is an authoritarian dictatorship under President-for-Life Emmanuel Macron. It is hostile to freedom, Enlightenment values, civil society, and rule of law. It controls all media, rewrites history to support its official narrative, and eliminates opposing viewpoints. France educates its people from birth to be blindly nationalistic, and restricts their information access to only government-approved sources. It’s eager to spread its controlocracy and totalitarian values to others, and has the economic might to do so.
Meanwhile, techno-optimists in the United States are excited about the trendy new technology of machine learning. What could be better for humanity than harnessing computers to tease patterns out of data? French tech companies are coming up with so many interesting business models, and developing innovative new technology, it’s so exciting!
Due to technological advances in machine learning created by the techno-optimists, France surveils, polices, and monitors its people cheaply and effectively. This allows France’s government to seize more resources from its people and use them in ways that violate human rights. For example, France uses these new technologies to establish and run a chain of concentration camps in Provence-Alpes-Côte d’Azur, where it imprisons the majority of the nation’s able-bodied Muslims.
In this nightmare scenario, the techno-optimists managed to grow the pie by developing new technology. They wrongly assumed that growing the pie leads to more prosperity. Growing the pie might be good, but not if the pie feeds people who oppose your values!
In nuclear technology, the dangers were symmetric. Nuclear weapons could be used to violate human rights, but they could also be used to defend human rights. That’s not the case with machine learning and the related cluster of technologies. These technologies will be used to curtail inviolable human rights.
This blind spot exists because techno-optimists use a conventional model of economics, where value is created and then shared between various parties. Implicit in these models is the assumption that the value captured by a party cannot be less than zero, and cannot exceed the value created. In situations like the machine learning scenario above, though, these constraints don’t hold. Value creation can indeed be harmful to the value creator, as we learned from the case of the Japanese POW slave laborers.
In the final reckoning, human rights trump economic growth. These asymmetrically bad technologies are especially dangerous, because they are alluring to first-order thinkers who calculate only the immediate benefits and costs.
 Economic growth compounds exponentially because improvements to existing processes, increases in productive population, and increases in capital all tend to be exponential functions.
 The full quote is, “We should make political choices so as to maximize the rate of sustainable economic growth, as defined by Wealth Plus.” Wealth Plus is “The total amount of value produced over some time period. This includes traditional measures of economic value, as would be found in GDP statistics, but also measures of leisure time, household production, and environmental amenities, as summed up in a relevant measure of wealth.”
 In the Solow model, technology is the only long-run steady state growth factor, because labor and capital are expected to reach a steady state equilibrium. In other models you’d also include the effect of institutions, culture, and so on.
 In Stubborn Attachments, a side constraint is defined according to Robert Nozick’s definition, based on the deontological principle of humanity.
 Open borders is another example of asymmetrically bad sociopolitical technology which is similarly alluring because it promises to increase overall prosperity, and is similarly attractive to first-order thinkers.